Page 1 of 1

signal-to-noise: standard calculation procedure?

Posted: Fri Feb 09, 2007 5:50 pm
by Kajjo
Hi everyone!
We are just discussing several different methods of calculating signal-to-noise ratios in chromatography (predominantly GC/MS).

baseline level = average value of local noise left and right of target peak inside 20*half-peak-width, while ignoring all data closer than 5*half-peak-width (20 and 5 = arbitrarily choosen)

noise =
a) difference between maximum and minimum value of local noise
b) standard deviation of local noise (a lot smaller!)

1) Do you know any formal requirements or standard procedures to calculate S/N ratios?

2) Do you determine the noise by method a or b? Or other? I know many people prefer standard deviation because it gives numerically better S/N. However, in spectroscopy (nmr) usually strict min-max values are used.

3) How large do you select the local noise range? How close, how far?

Obviously, the S/N is only valid for a particular acquisition method, because scan rate, scan range, dwell time and so on do influence the S/N ratio. Naturally, spikes are part of the noise. Only reproducible peaks may be ignored.

Thanks for your comments!

Kajjo

Posted: Mon Feb 12, 2007 1:02 pm
by HW Mueller
John W. Dolan, LC.GC Europe, January 2006, p. 12.

Posted: Fri Feb 16, 2007 11:54 am
by Kajjo
Thanks for the reference. However, my questions are not addressed at all in that article. It is just very basic.

Kajjo